Objects detection and recognition in smart vehicle applications: Point cloud based approach

摘要

In this paper we describe our experience and a progressing work in investigating the embedded intelligence in smart automobile. First, we present a new smart embedded computing algorithm for objects detection and recognition. This algorithm combine two important tasks, Segmentation and Box-Slicing. And then, we validate our software algorithm using a LIDAR point cloud data as sensor input on a real HW/SW embedded architecture based on ZedBoard FPGA. The hardware architecture is built on the top of two processor ARM, in the other hand the software application is built around an embedded Linux/ Linaro distribution. This work is considered as a first and important step toward the reconstitution of automobiles environment.